Skip to content

Commit cd28f54

Browse files
committed
Fix for 32 bit part 2
1 parent e5d987c commit cd28f54

File tree

1 file changed

+8
-8
lines changed

1 file changed

+8
-8
lines changed

src/TriangularSolve.jl

Lines changed: 8 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -86,15 +86,15 @@ end
8686
# @inline function solve_Wx3W!(ap::AbstractStridedPointer{T}, bp::AbstractStridedPointer{T}, U, rowoffset, coloffset, m::VectorizationBase.AbstractMask) where {T}
8787
# WS = VectorizationBase.pick_vector_width(T)
8888
# W = Int(WS)
89-
# A11 = vload(bp, Unroll{2,1,W,1,W,0xffffffffffffffff,1}((rowoffset,coloffset)), m)
90-
# A12 = vload(bp, Unroll{2,1,W,1,W,0xffffffffffffffff,1}((rowoffset,coloffset+WS)), m)
91-
# A13 = vload(bp, Unroll{2,1,W,1,W,0xffffffffffffffff,1}((rowoffset,coloffset+WS+WS)), m)
89+
# A11 = vload(bp, Unroll{2,1,W,1,W,(-1%UInt),1}((rowoffset,coloffset)), m)
90+
# A12 = vload(bp, Unroll{2,1,W,1,W,(-1%UInt),1}((rowoffset,coloffset+WS)), m)
91+
# A13 = vload(bp, Unroll{2,1,W,1,W,(-1%UInt),1}((rowoffset,coloffset+WS+WS)), m)
9292

9393
# A11, A12, A13 = solve_Wx3W(A11, A12, A13, U, WS)
9494

95-
# vstore!(ap, A11, Unroll{2,1,W,1,W,0xffffffffffffffff,1}((rowoffset,coloffset)), m)
96-
# vstore!(ap, A12, Unroll{2,1,W,1,W,0xffffffffffffffff,1}((rowoffset,coloffset+WS)), m)
97-
# vstore!(ap, A13, Unroll{2,1,W,1,W,0xffffffffffffffff,1}((rowoffset,coloffset+WS+WS)), m)
95+
# vstore!(ap, A11, Unroll{2,1,W,1,W,(-1%UInt),1}((rowoffset,coloffset)), m)
96+
# vstore!(ap, A12, Unroll{2,1,W,1,W,(-1%UInt),1}((rowoffset,coloffset+WS)), m)
97+
# vstore!(ap, A13, Unroll{2,1,W,1,W,(-1%UInt),1}((rowoffset,coloffset+WS+WS)), m)
9898
# end
9999

100100
# solve_3Wx3W!(A,B,U::UpperTriangular) = solve_3Wx3W!(A,B,parent(U))
@@ -226,15 +226,15 @@ end
226226
quote
227227
$(Expr(:meta,:inline))
228228
# here, we just want to load the vectors
229-
C11 = VectorizationBase.data(vload(spa, Unroll{2,1,$W,1,$W,0xffffffffffffffff,1}((StaticInt(0),n)), mask))
229+
C11 = VectorizationBase.data(vload(spa, Unroll{2,1,$W,1,$W,(-1%UInt),1}((StaticInt(0),n)), mask))
230230
Base.Cartesian.@nexprs $W c -> C11_c = C11[c]
231231
for nk SafeCloseOpen(n) # nmuladd
232232
A11 = vload(spc, (MM{$W}(StaticInt(0)),nk), mask)
233233
Base.Cartesian.@nexprs $W c -> C11_c = vfnmadd_fast(A11, vload(spu, (nk,n+(c-1))), C11_c)
234234
end
235235
C11 = VecUnroll((Base.Cartesian.@ntuple $W C11))
236236
C11 = solve_AU(C11, spu, n, Val{$UNIT}())
237-
i = Unroll{2,1,$W,1,$W,0xffffffffffffffff,1}((StaticInt(0),n))
237+
i = Unroll{2,1,$W,1,$W,(-1%UInt),1}((StaticInt(0),n))
238238
$storecexpr
239239
maybestore!(spb, C11, i, mask)
240240
end

0 commit comments

Comments
 (0)